19 November 2012

Resume

This weblog has, unfortunately, been idle for a very long time. Between work and my kids, I just haven't had the time to put into it. It does still show up on Google, though, so I thought I'd take the opportunity to put up my resume. If you are an employer in the DC, Chicago, Twin Cities, or Denver area who is looking for a spectroscopist or optical materials expert, please drop me a line.


Nicholas J. Condon, PhD
Objective:
A challenging position in a research and development environment
Qualifications and Skills:
·            Expert lasers and optical materials with 15+ years lab experience
·            Lasers, laser design, and optical cooling
·            UV/Vis, NIR, and mid-IR spectroscopy of organic and inorganic materials
·            Optical materials synthesis, purification and doping
·            Thermal modeling and measurement of optical materials
·            Nonlinear optical materials, nonlinear optics and spectroscopy
·            Anhydrous and controlled atmosphere work
·            Mathematical modeling
Work Experience:
2001-Present: Research Chemist; Optical Sciences Division, US Naval Research Laboratory; Washington, DC (2001-2004 as a contractor through SFA, Inc., Largo, MD; government employee thereafter)
·         Successfully pitched a M$1.1 program on halide nonlinear optical materials as its principal investigator
·         Developed improved crystal growth methods for laser materials critical to the success of two million-dollar programs
·         Synthesized, purified, doped, and grew single crystals of halide-based laser and nonlinear optical materials
·         Developed dissolution and spectroscopy methods for purity and dopant analysis of optical materials
·         Testing and modeling of new nitride optical materials
·         Made world-first measurement of optical cooling in an erbium-doped laser material; work was recognized with an NRL Alan Berman Research Publication Award
·         Modeled optical processes and heat flow for comparison to data
·         Collaborated on the design and construction of solid-state lasers in spectral regions ranging from the blue to the mid-infrared, including kW-scale systems
·         Multiple research publications, conference presentations, and patent applications

1996-2001: Graduate Assistant; Department of Chemistry, University of Wisconsin; Madison, WI
·         Measured vibrational coupling between modes of organic nitriles using a new nonlinear laser spectroscopy technique
·         Studied the infrared and Raman spectra of fundamental and overtone bands in organic nitriles
·         Modeled the linear and nonlinear spectroscopic responses of nitriles in MATLAB using density matrix theory
·         Designed, constructed, and tested a dispersive Raman spectroscopy system
·         Education in and experience with GC, HPLC, ICP, AA, electrochemistry
·         Multiple research publications and a conference presentation
1995-1996: Research Assistant; Climate Monitoring and Diagnostics Laboratory, National Oceanic and Atmospheric Administration; Boulder, CO
·         Analyzed air samples for ozone-depleting compounds via gas chromatography/mass spectrometry (GC/MS)
1993-1994: Research Assistant, Basic Sciences Center, National Renewable Energy Laboratory, Golden, CO
·         Assisted with inorganic synthesis under controlled atmosphere
·         Assisted in culture preparation and extraction in microbiology laboratory
Education:
2001 - PhD, Analytical Chemistry; University of Wisconsin; Madison, WI
Thesis:  "Doubly Vibrationally Enhanced Infrared Four-Wave Mixing Spectroscopy of Acetonitrile and Crotononitrile"
Advisor:  Prof. John C. Wright
1996 - BA, Chemistry and Mathematics; University of Colorado; Boulder, CO
Publications:
20 peer-reviewed publications, 25 conference presentations (8 as presenter), 1 patent, 2 patent applications (full list available on request)
Awards, Honors, and Other Accomplishments:
Three-time champion on the television game show, Jeopardy!, June 2011
Alan Berman Research Publication Award; Naval Research Laboratory, 2010
Roger J. Carlson Award; Univ. of Wisconsin, Dept. of Chemistry, 2000

02 October 2011

Color Photography

My Jeopardy! winnings arrived recently. I'm donating a bunch of the money to various charities, and a bunch more is going to savings and to improvements to our house, but I have allowed myself a few indulgences. One of these indulgences is a professional-grade monitor, am NEC MultiSync PA231W. In addition to making everything look much better than it does on an ordinary monitor (seriously, you would not believe the difference), the big advantage to this monitor is that is is color correct; that means that I can now have a high degree of confidence that what I see on my monitor will reflect what will show up in a print. This makes the prospect of doing color photography a lot more appealing. And, while I do love monochrome photography above all things, I have found a few photos recently that are better in color. People seem to respond very well to color photography, so perhaps this gives me an opportunity to show people something that they might appreciate more than my usual black-and-white.


The propellors above were photographed at the College Park Aviation Museum, which is suprisingly interesting given its location. While it hardly holds a candle to the Smithsonian, I'd strongly recommend it to aviation buffs who've already seen everything that Air and Space has to offer.


I titled the above photograph, "Lake Vacation Cliche." I am exceedingly pleased with the photo, but I've seen photos of a boat reflected in a lake from dozens of other photographers. As much as I like the picture, it's clearly been done before.


The above photo was taken in Annapolis. The colors have been bumped up a bit from how they look in reality, and, based on my Flickr comments, it's clear that the result is quite popular. I haven't printed very many of my color photos in the past, but perhaps I'll print this one just to see what it looks like.

25 September 2011

Jeopardy! ToC Odds Update: New Season, New Superchamp

So, Justin Sausville (a former student of one of my neighbors, interestingly enough) has finally lost after an oustanding run of six games. The Tournament of Champions field now has four superchamps.

Given that I am now receiving some gentle elbows to the ribs on the subject of updating these rankings...

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Justin Sausville6 Wins$134,000100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,411100.0%100.0%-100.0%
Jay Rhee5 Wins$107,203100.0%100.0%-100.0%
Kara Spak5 Wins$83,401100.0%100.0%-99.8%
Mark Runsvold4 Wins$153,80099.9%100.0%-99.5%
John Krizel4 Wins$105,20499.0%99.8%-97.2%
Brian Meacham4 Wins$90,50091.7%96.7%-84.8%
Buddy Wright4 Wins$88,80471.2%83.3%-59.3%
Sara Heard4 Wins$85,60133.6%47.0%-24.0%

Note: Thanks to a comment from indyjeff, I have revised the numbers to reflect the correct ToC Start Date of 2 November 2011.

So, John Krizel and above remain near-locks, Brian Meacham is very likely to make it, Buddy Wright is in a pretty strong position to make it, and Sara Heard will more likely be the alternate than a competitor.

All condolences to Paul Wampler, who is an excellent player denied a spot in the ToC by his misfortune to have competed in a particularly brutal ToC cycle.

As I understand it, the ToC starts filming on Tuesday, so an amusing fact about all of this is that I'm predicting something that's already set in stone. Congratulations to everyone who made the final field (whoever they may be), and I look forward to watching you play in a few weeks!

23 July 2011

Jeopardy! ToC Odds: Big Money Edition

Well, Mark Runsvold has lost after a short but spectacular run of four games that put him among the highest-earning Jeopardy! players ever. Congratulations to him!

Of course, this means that it's time for a new odds update:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,411100.0%100.0%-100.0%
Jay Rhee5 Wins$107,203100.0%100.0%-100.0%
Kara Spak5 Wins$83,401100.0%100.0%-99.8%
Mark Runsvold4 Wins$153,80099.7%100.0%-98.8%
John Krizel4 Wins$105,20498.4%99.7%-95.0%
Brian Meacham4 Wins$90,50089.4%96.6%-78.7%
Buddy Wright4 Wins$88,80471.0%89.3%-54.1%
Sara Heard4 Wins$85,60142.7%61.9%-28.3%
Paul Wampler4 Wins$72,00112.1%22.7%-6.5%

For most of the folks on the table, very little has changed. The 5+ win crowd still have nothing to worry about, and the odds of the four 4-timers below Mark haven't moved much at all; it's clear that the appearance player at about Mark's ranking was sort of "priced in" to the odds. The only two people to suffer much in this most recent calculation are Paul Wampler, who now needs a complete drought of 4-timers above him to make the tournament, and Anthony Fox, who has been pushed off the bottom of the board entirely. Sorry, Anthony; I was rooting for you.

The most likely field right now is Buddy Wright and up, along with two yet-to-play contestants and Sara Heard as the alternate. Almost as likely is a field of Sara and up, with one yet-to-play and Paul as the alternate.

28 June 2011

Jeopardy! Toc Odds: New Update

Congratulations to DC local guy Jay Rhee for his 5-game winning streak. With his addition to the list, the odds have changed again:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,411100.0%100.0%-100.0%
Jay Rhee5 Wins$107,203100.0%100.0%-99.8%
Kara Spak5 Wins$83,40199.8%100.0%-98.7%
John Krizel4 Wins$105,20498.2%99.8%-93.8%
Brian Meacham4 Wins$90,50089.1%97.2%-75.9%
Buddy Wright4 Wins$88,80472.8%89.3%-54.1%
Sara Heard4 Wins$85,60149.3%71.8%-31.0%
Paul Wampler4 Wins$72,00120.3%38.6%-10.2%
Anthony Fox4 Wins$51,9983.2%8.4%-1.2%

So, Jay Rhee's addition has shifted the field somewhat; Sara Heard is now squarely centered on the bubble position. The likeliest field features Buddy Wright on up and three players yet to be named; almost equally probable is a field consisting of Sara Heard on up and two additional players. Paul Wampler is about even odds to occupy the alternate's spot, and Anthony Fox is likely going to be on the outside looking in.

This ranking also marks the final elimination of the last three-time champion; barring a highly improbable series of events, this year's ToC will consist entirely of winners of at least four games.

With the Jeopardy! boards offline for the foreseeable future, I have no way to announce to interested parties that this list has been updated. So, if you know anyone (like former boardies) who might be interested in this information, please be sure to pass it along to them!

18 June 2011

My Jeopardy! Story: Game 4

With this game, my run on Jeopardy! ends, one win shy of making it to the Tournament of Champions.

From start, this game was a fight. I had Stewart pegged as dangerous from early in the day; he was confident, calm, smart, and fast. Lisa, despite being quiet, turned out to be a pretty tough opponent herself. The three of us traded clues through the first round, with the both of them outbuzzing me on a very regular basis. I caught the DD, but I only had $1400 in the bank, so my True Daily Double wager was a lot less satifying than my one in the previous game. My struggles with the "Fun With Elements" category in this round were not my among proudest Jeopardy! moments.

We entered Double Jeopardy! in a fairly close game. I had a good run on the buzzer to start the game, enhanced by picking up a couple of rebounds off of Stewart, whose brain went blank a couple of times. (Stewart's a foreing service officer, he damn well knows, "Kabul.") I found the first DD with a huge lead and proceeded to underwager $4000, since I already had a lock game and the extreme vagueness of the category scared me a bit; I should have wagered something more like $8000. From there, though, Stewart simply took over the game. I got in on three of the remaining 15 questions, missing one of them, Lisa hit two of them, and Stewart took the rest of them. He made a couple of big misses, a couple of big hits, and finally, with four clues remaining, found the last Daily Double. (In a bit of irony, the correct response to the last clue before the reveal of the DD was "awkward," the most appropriate adjective to describe my camera appearance; I was outbuzzed by a tiny fraction of second.) Steward still trailed me, with about $11,000 to my $17,000, so he made exactly the correct strategic move: He wagered $10,000. He provided the correct response to the clue, and took the lead from me. Stewart's DD hunting had depleted all of the high-value clues, so what remained was insufficient to let me recover the lead, and for the first time, I entered Final Jeopardy! in second place.

As I was making my Final Jeopardy! wager, I noted that I was four-fifths to Stewart, so I very carefully wagered $9799. This amount made it so that I'd win no matter what if Stewart made a rational covering wager and he missed. It was, in essence, the largest wager I could make in this case; I could easily have wagered less. Although thought out carefully, was a huge mistake.

I completely forgot to take Lisa into account. She had $5600, and had she wagered everything and responded correctly, she would have had $11200. As this is considerably more than the $8401 I'd be left with in the event of a miss, she would have won handily had Stewart and I both failed to respond to the Final Jeopardy! clue correctly. Oops.

As it turned out, the wager didn't matter. Though I managed to come up with a non-idiotic response to Final Jeopardy!, it was incorrect, while Stewart gave the correct response. (Lisa, FWIW, also missed.) I took second place.

There were many places in the game where a small change could have won it for me. Had I outbuzzed Stewart and answered a couple more high-valued clues, I would have held my lead even against his huge DD wager. Had I gotten in on the clue immediately before the DD, I would have selected the DD myself and put Stewart away with a $6000 bet.

Of course, aliens could have landed on the set and proclaimed me Prince of the Galaxy; that didn't happen, either.

Jeopardy! was a fantastic experience, though. I fulfilled a lifelong ambition by becoming a Jeopardy! champion. I met some interesting people on the set, and got to know many more interesting people as I joined the Jeopardy! community online. I earned enough money to buy some improvements and furniture for our new house, as well as a new camera and some other photography stuff. While I will be forever a bit heartbroken about losing that last game, it was, taken as a whole, kind of awesome.

16 June 2011

Jeopardy! ToC Odds: The "Conspicuous Absence" Edition

So, given that it's been a while, it's time for a new ToC odds update:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,411100.0%100.0%-100.0%
Kara Spak5 Wins$83,40199.9%100.0%-99.4%
John Krizel4 Wins$105,20499.3%99.9%-96.7%
Brian Meacham4 Wins$90,50094.2%98.9%-84.2%
Buddy Wright4 Wins$88,80483.7%95.3%-66.9%
Sara Heard4 Wins$85,60166.1%85.7%-45.7%
Paul Wampler4 Wins$72,00137.1%60.9%-20.6%
Anthony Fox4 Wins$51,99811.0%24.8%-4.6%
Megan Barnes3 Wins$103,2031.7%*5.2%-0.6%
Alison Stone Roberg3 Wins$85,1021.6%4.8%-0.5%

So, the most probable field is Sarah Heard and above, plus three people whose episodes have yet to air, with Paul Wampler as the alternate.

As was widely predicted, I did originally develop this tool to figure out my chances at getting into the ToC. In the beginning, before Tom Kunzen played, it told me that my chances were poor. As the weeks passed, my chances slowly declined to zero. With 3 wins and $80,101, I enter the list in 17th place, out of contention. I will not be going to the ToC.

That won't stop me from continuing to tend this odds calculation, though; I wasn't lying when I said that I was just the kind of geek who would do this for fun. So, while I won't be there myself, I will continue to remain interested in who will be.

My Jeopardy! Story: Game 3

My third game was the sort of game I'd always wanted to play. This one was very satisfying.

After two games on stage, the routine of getting set up had become familiar. While my physiological stress indicators were still at the rails, I felt less nervous. I had proven that I could win, so now I could play more for fun.

After a little bit of noodling around at the beginning of the first round, I started to lock in on the buzzer and hit clues that I could answer. As in my previous game, I was playing the first round straight rather than bouncing, and things went swimmingly. After taking three of four (and getting outbuzzed on the one I didn't get) in what turned out to be a pretty easy category about Houdini, I hit the Daily Double on the $1000 clue. I had $5000 in the bank and a good, solid lead, so it was time to put my money where my mouth was: I made it a true Daily Double. It was both fun and exhilarating to say it, and it seemed to set Alex back a step. The clue asked for the mystery novelist and gullible naif advocate of spiritualism that was a former friend of Houdini's. "The dude who wrote Sherlock Holmes" popped immediately into my head, and it took my stress-addled brain a second to convert that to "Arthur Conan Doyle." The big wager had its intended effect: It opened up a big lead, and it tilted my opponents somewhat; neither of them answered much for the remaining half of the round.

The Double Jeopardy! round started somewhat inauspiciously for me, with Michael beating me to the buzzer four consecutive times in the "Bands by Lead Singer" category and eating up a hunk of my lead. This put The Fear into me a bit, but it turned out to be Michael's high water mark, he wasn't a factor in the rest of the game. I took several of the next eight clues, then hit my first DD under the $1600 space in "Weather, Man." I looked at my lead and wagered $3000, enough to put the game in lock territory. Looking back on it, this seems like a failure of aggression on my part; my lead, which I had held since the fourth clue, had made me complacent. $6000 would have been a much better wager, especially given that meteorology, as a science category, is edging into wheelhouse country. As it turned out, I got the (fairly easy) clue correct, picked up my $3k, and never dropped out of lock territory for the rest of the game.

Another sign of my complacency here, as has been pointed out to me by a scholar of such matters, was that I didn't make any effort to hunt the second daily double. This almost caused me some difficulty, since just a few clues later, John found it. He had $7000, and, with admirable strategic aplomb, wagered $6000 of it. Fortunately for me, he missed the correct response (which I would have missed, also, FWIW) and took himself out of the game completely. Had he nailed it, my lock would have been gone, and I would have needed Final Jeopardy! to seal the win.

As it stood, though, these two bits of insufficient aggression didn't hurt me. I ended the game with a solid lock and wagered $5000, near my maximum Clavin's Rule bet, on "Crusading Women." I wasn't thrilled about the category, but I wanted a shot at another $30k game. The clue came: "A judge's directed verdict of guilty for her action in Rochester in 1872 was written before her trial began." I had no clue of the specific incident referenced, but the date told me I was probably dealing with a suffragette, not a temperance crusader. "Name a suffragette" is a query that should return "Susan B. Anthony" as the first hit, so that's what I went with. I was correct.

I finished my third game, the best I'd played so far, with $30,200, a dollar less than my first day total. My overall total was now $80,101. With that, it was off to lunch and Game 4...

15 June 2011

My Jeopardy Story: Game 2

Game 2 turned out to be a lot closer than I remembered and a lot closer than the final score indicated.

As I was brought out to the stage for Game 2 was calling me, "Champ." It was a strange feeling. I stepped up to the champ's podium on the right, with LuEllen taking the middle and Jeff taking the left. We wrote down our names, tested our buzzers (we did this before every start or restart of play, so there could be no question that they were working properly), the theme music played, and we were off and running.

As much to get the feel of it as anything, I decided to try playing the first round "straight," with no bouncing. I figured (correctly, as it turns out) that the other two would be unlikely to make the strategically-correct putaway bet if given the opportunity on a DD, so I wasn't as worried about finding it.

I started stronger than I had in the previous game, but the board was much less favorable to me. I beat LuEllen to the buzzer much more often than not when I wanted to ring in, but she went after a lot of questions that I just didn't know.

I came out of the first round with $7000 to LuEllen's $5000, with Jeff a non-factor in third. Jeff started things off, then LuEllen started gobbling down questions in the "Art" category. Categories about art and sculpture are among my very worst, and I didn't even try for the buzzer before LuEllen hit the first DD. Trailing by $2000 in what was clearly a wheelhouse category, she bet $2000. Looking back on it, I strongly disagree with this play; she should have taken a true DD and surged into the lead. At the time, I was quite happy to see a conservative wager. She got the question (the only one in the category that I could have answered correctly) and play continued.

A few clues passed without me getting in, then LuEllen hit the second DD. She wagered $2000 again (a more defensible wager in a random category like "'Go,' Baby, Go"), and was asked a geography question that I couldn't have answered in a million years. She couldn't answer it either, and with that, both DDs were gone with zero net effect on the score.

From there, I was able to start making some progress; my score steadily rose past LuEllen's. Jeff, though, went on a surprise tear, rocketing from a distant third all the way up to a few hundred out of second place. I finally put a stop to it and ran off a few of my own.

With one question left, $2000 in the category, "Great Lakes" (famous people with "Lake" in their name), I had $17800, LuEllen had $9800, and Jeff had $9200. I glanced at the scores very quickly and realized that I needed to ring in on the last clue no matter what; if I got it right, I'd lock the game, and if I got it wrong, I'd still have a crush over both of the other contestants. I set myself to ring in and the clue was revealed; it amounted to, "What football position is designated, 'SS'?" I knew the answer was "Strong Safety," and when I saw the lovely red lights, I knew I had won.

I had always wanted to win a game of Jeopardy!, and I did so in Game 1. My next goal was to lock a game, and I did that in Game 2. It felt great.

I made an itty-bitty wager of $100, and the Final Jeopardy! clue about flags popped up: "'L'unifolie' is one of the names popularly given to the new flag unveiled in 1965 by this country." I knew they wanted a French-speaking country, but, for some reason, my brain wouldn't translate the French. (I don't speak French, but my mom's a French teacher, and I darn well should have been able to see the cognates there.) So, I tried to think of a former French colony in Africa, came up blank (Nice going, brain! Thanks for that!), and flailed with "Haiti." That was wrong, of course, since on Trebek's Jeopardy!, the answer is always "Canada."

So, as it turned out, I was damn fortunate to finish my lock. If one of the other two beat me to the buzzer on that last question, LuEllen would have beaten me and I'd have gone home singing, "Blame Canada!" Instead, I won $19,700, bringing my total to $49,901, and it was back to the Winner's Circle interview and the green room to get ready for Game 3...

14 June 2011

My Jeopardy Story: Game 1

My second day on Jeopardy! started as a nearly-exact duplicate of my first: Shuttle bus ride to the studio, security screening, paperwork, high-intensity briefing, practice session. The only differences were that I already had my Hometown Howdy ready, that I felt a lot better in practice, and that I was selected to play in the first game of the day.

That last bit sent my pulse up to a tempo that a classical musician might refer to as "prestissimo" and that a rock musician might refer to as "speed metal."

My makeup got a touch-up, the sound guy clipped a microphone on me, and I was off to the stage. I was given the center podium, with Amy, the returning champ, on my right and Daniel on my left. I had seen Amy play the previous day, so I knew that, despite her very evident nerves, she could hang on for the win in a tough game. Daniel worried me a bit more; he was very strong on the buzzer in practice, and seemed quite confident.

Before discussing the gameplay, I will note that, before watching, I remembered exactly two questions from the game: The first Daily Double in Double Jeopardy!, and Final Jeopardy! The rest of the time, I was so immersed in what I was doing that I did not have the spare brain time to write memories. A lot of it came back to me as I watched the game, though.

While I got the first question right (a fact that my father remembered, but I didn't), the round started very slowly for me. I knew most of the answers for the first 15 clues, but I struggled to outbuzz Daniel. I was not pleased to go into the first commercial with $200 on the board.

After the commercial (and the awkward chat), things went much better. I started into some wheelhouse categories for me, such as the food one, and I was able to find my swing on the buzzer.

Ringing in on a Jeopardy! question is an incredibly intense experience. The clue is read, there is a lag (in which empires could rise and fall, glaciers advance and retreat, species evolve and become extinct...), then the lights come on. I almost always knew if my timing was right the moment I hit the buzzer, and the red podium lights on the bottom of my peripheral vision that signalled my success always produced a tiny surge of extreme joy and relief. There are few things in my experience that compare to it.

I ran off half a dozen or so correct responses in a row, and closed most of the gap with Daniel. He then hit a couple of questions followed by a Daily Double. He wagered $2500 (too small, in my opinion, but better than the $1000 wager to which so many contestants default), and got it right. By the end of the first round, he was still ahead, but only on the strength of that DD. Between the two of us, though, we ate up almost all of the clues and left Amy well behind.

A note on my clue selection: In defiance of traditional Jeopardy! gameplay, I did not select the clues in order, starting at the top and moving down. Instead, I picked third-row or below clues and switched categories often in a strategy called the Forrest Bounce. This has proven to be effective for some people, since it disorients your opponents a bit and increases your chances of hitting a Daily Double. It is not popular with many at-home viewers, and Alex and the producers just despise it. I thought I'd give it a shot anyway.

Double Jeopardy! went even better for me than the first round di. Daniel was getting some clues, but I was outbuzzing him more often than not, and Amy continued to struggle to get in. I pulled out to a good lead, then hit my first Daily Double. I wagered $5000, which was enough to give me a very large lead if I got it correct, but wasn't so much that I'd drop behind Daniel if I got it wrong. I didn't remember the exact wording of the clue, but I knew that the answer was William of Ockham the moment I read it. As a hardcore skeptic, Ockham's Razor is one of the central tenets of my entire belief system; I could hardly have been given a better clue, and from the reaction of the folks at my viewing party, a lot of my friends knew that.

It's a classic mistake in Jeopardy! to give someone's full name when only their last name is needed (since you might botch the first name and render your response invalid), so I responded "Who is Ockham?" Alex said "Yes, William of Ockham" and the game proceeded.

I picked the next clue and hit another Daily Double. Taken aback, I wagered $5000 again, then the game was stopped by the producers and we were told to turn our backs to the board.

Game stoppages like this happen every so often, but the recording is always edited to conceal them from home viewers. Sometimes they happen for technical reasons, but most of the time they happen because a player has given a debatably correct answer, and the staff needs to decide whether it should be counted as correct or not. While I was never formally told the reason for the delay, I strongly suspect that it was over my response of "Ockham" when the correct response was "William of Ockham." To my mind, the former is entirely correct under the scoring rules of Jeopardy!; since there is no one else of any note who could be readily identified by the name, "Ockham," I did not need to be more specific than that.

Apparently, the producers agreed. The tape was rolled back to just after I selected the clue, and I was told that, since no scores were changed, I was required to restate my wager of $5000 during the re-recording. The clue was a Video Daily Double, so one of the Clue Crew took longer than necessary to ask for the names of the two bones in the forearm. I thought about it for a second to make certain that I was giving the bones in the arm rather than the leg, then replied, "What are the radius and the ulna?" This correct response put me far ahead of Daniel (~$27,000 to Daniel's ~$12,000, if I recall correctly), and I hoped it would seal a lock game for me.

Unfortunately, Daniel was unrattled by this and he proceeded to rack up a number of correct responses. I laid back, since I wasn't 100% sure of the correct response on many of the clues, and I didn't want to Jeopardize (Hah! See what I did there!) my lead by guessing. With three clues left, totalling $1600, I was $2000 from a lock. I answered the last three, even though I knew I couldn't lock the game, and it was off to Final Jeopardy! with me at $29,800 and Daniel at $15,100.

The category was "Show Business Families;" this is pretty close to a wheelhouse category for me, since I'm so strong in TV and movies. I knew that I needed to wager a minimum of $401, but I could wager much more aggressively than that if I so chose. Given my strength in the category, I briefly thought of making a big wager, like $10,000. I then thought about whether, in other circumstances, I would be willing to wager $10,000 of my own money on a single trivia question. A Jeopardy! score is just a pile of valueless points until you win, but if you win, it becomes real money. I needed to wager $401 to have a shot at winning; I did not need to wager enough money to replace the horrible, 60-year-old single-pane casement windows on our recently-purchased house, or to buy a complete dining room set, or to take a trip to Europe. I wagered $401.

The clue asked for the name of a director whose middle name was that of a car company. I immediately started flipping through the long list of directors in my head, looking for a likely middle name. This went on for about five seconds before a wiser part of me said, "You idiot! You know hundreds of directors, but there are only a dozen or so car companies! Chevrolet: No. Chrysler: No. General Motors: Are you frakking kidding me? Ford: Wait, yes, it's Coppola's middle name!" I wrote down Francis Ford Coppola, and was all but certain that I had won, but I didn't relax yet.

Amy's response was revealed; she also wrote down Francis Ford Coppola, and was ruled correct. I finally relaxed. Daniel got it correct as well, but the look on his face suggested that he knew I had gotten it, too. After my reveal, I was told that I was a Jeopardy! champion, and that I had won $30,201.

I had a very few seconds to let this fact sink in, then my mic was unplugged and I was directed out onto the floor to chat with Alex while the credits rolled. He spent the time criticizing me for using the Forrest Bounce, and encouraged me to play straight next time so that I could get into the rhythm of the categories and not throw myself off. I found this criticism both annoying and misplaced, given the long runs I had in both rounds.

With that concluded, I was whisked off to the green room to change clothes, use the bathroom, guzzle down a can of Diet Coke, and (whimper) write a new Hometown Howdy. Half an hour later, I was back out on stage for Game 2...

12 June 2011

My Jeopardy! Story: On Buzzers

Before going out to California for my taping, I read everything I could find on the internet about buzzer strategy. There seemed to be a uniform consensus as to the proper technique: Listen to Alex's voice, and try to match your buzzing rhythm to that of the production staff member who unlocks the buzzers when Alex is done reading; never wait for the lights, or you'll be too late. Although this can obviously be effective, my own experiences and observations suggest that this advice can be problematic for some people.

As I discussed in my previous post, there are things about the Jeopardy! set that are not at all apparent to the home viewer, or even to most of the studio audience. One of these things is a particular monitor on the table where Maggie, Robert, and Corina, the contestant coordinators, sit. (I don't remember if Glenn sits there or not; he just seemed to sort of materialize when needed.) This monitor is not visible at all from most of the studio audience, and is never shown on television. It is, however, clearly visible from the seats where waiting contestants sit. Said monitor shows a live view of the contestants, but more importantly, it also shows their buzzer status. All three buzzers are represented with three indicators: one that shows when the button is being pressed, one that shows when the contestant is locked out (from buzzing too early), and one that shows when a contestant has successfuly "locked in" and earned the chance to answer the question. I noticed this monitor while watching the second game on my first day, and spent much of the next three games watching it closely. The conclusion from my observations is that many players who struggle with the buzzer, especially the younger or more nervous ones, have a terrible problem with ringing in early.

My experience in the practice sessions on my first day suggest that I am among those not well-served by the conventional wisdom on ringing strategy. I waited for Glenn (in his role as the Alex of the practice sessions) to finish the clue, then timed my ring for what seemed like a reasonable period of time afterward. This led to me getting outbuzzed half a dozen times. So, I extended my wait period. I waited until it felt as if an eternity elapsed after the end of the clue before I pressed the button; this technique got me in a couple of times before I was rotated out. My second round of practice was a little better, but there was no doubt that the buzzer was going to kill me unless I changed tactics.

For the third round of practice, after lunch, I decided to discard the conventional wisdom and wait for the light. As a twitchy, high-stress, highly-caffeinated individual who has logged several thousand hours playing reflex-intensive action video games, I figured that my reaction time might be fast enough to make this work. So, I adopted a new strategy: When a new clue was revealed, I'd read the clue, decide if I was going to answer, then ignore Glenn and focus on the light. This yielded immediate results, as I got in several times in quick succession. Maggie, seeing that I'd found my swing, told me to sit down. The next morning's practice sessions netted similar results, so I felt good about my new technique going into real competition?

How did this work out for me? Tune in tomorrow to see...

11 June 2011

My Jeopardy! Story: The First Day

TL; DR: I did not get to play Jeopardy! on my first day as a contestant. I watched other people play, and learned a great deal about the show.

---

Despite the appearance the producers like to present, Jeopardy! is not a live television show. It tapes in advance (in my case, four months in advance) and on a compressed schedule. Typically, they tape five episodes a day for two consecutive days. Each shooting day requires ten new contestants, since each ep has two new contestants and one returning champ. The producers, however, do not want to risk bringing in exactly ten contestants on each day, since one contestant may not show up or fall ill or otherwise become suddenly unavailable. So, on the first day of shooting, they bring in 13 people: One returning champ, ten new players from out of town, and two L.A. area locals who had previously been alternates. The two L.A. locals are guaranteed to play, while two of the ten out-of-towners will (normally) sit in the audience all day and be held over for the second day of taping. They draw the contestants for each game randomly (under the supervision of an independent attorney) before each game, so no one knows when they will play until the conclusion of the previous episode's taping. On the second day, there are again 13 people: One returning champ, eight new out-of-towners, the two carryovers from the first day, and two L.A. locals as alternates. The two carryovers are guaranteed to play (usually in the first two games) and the L.A. locals won't play unless needed.

As it turns out, having a day to sit in the audience was great. I would have enjoyed it more had I known I would not get to play that day, but it was still fascinating and valuable.

The day started with a shuttle ride from the hotel to the studio, followed by a trip through a metal detector and into the Jeopardy! green room. (N.B. The room is actually painted in a sort of pale mint green; I do not believe this is mandatory for green rooms in general, though.) We helped ourselves to some pastries and beverages, did some paperwork under the direction of Robert and Corina, and were instructed to begin thinking about our "Hometown Howdies."

"Hometown Howdies" come in two flavors: simple, and painful. The former variety, as my name for it would imply, is easy: "Hello, Lakewood! This is Nick Condon. Watch me on Jeopardy!" The latter variety requires writing and delivering some sort of joke or witticism specific to the town or region of one's origin. The result is all but universally awkward. Not my favorite part of being on the show.

Maggie then showed up and, for a loud, energetic, and entertaining hour, gave us a combination of an orientation and a pep talk. Meanwhile, we all rotated through the makeup  chairs. I was informed that I would look too pale next to the well-tanned Alex, so the makeup artists adjusted my skin tone appropriately. I'm pretty sure it was the first time I've ever worn makeup of any sort.

Finally, we were brought out to the main stage. The geometry of the stage is a bit different than it looks on TV. The full-stage shots must be taken with a very wide-angle lens, because the angle between the board and the row of podiums is considerably less than 45 degrees, while it appears to be greater than 90 degrees from my perspective on the couch.

I counted six cameras: Two just to the left of the board for shooting the playing contestants (I believe one provided the view of all three contestants while the other would swing to whoever was responding), a boom camera and two cameras on dollys in front of the stage to shoot Alex and the questions, and one fixed camera, mounted high above the entryway, for shooting the whole board. Above the cameras to the left of the board are the current scores of each contestant (note the direction in which a contestant presented with a Daily Double looks before making a wager), and below them is a large monitor for the video clues. The contestant coordinators sit at a table to stage left, while the rest of the production staff (including the dispute resolution group and the guy that unlocks the buzzers) sit at a much longer table to the center and right. The audience (of about 200?) sits in banks of seats that strongly resemble those found in movie theatres.

We were brought up to the podiums for an in-person briefing on how to use the buzzers and how to write our names and Final Jeopardy! responses on the lightpen screens. The podiums look from the back much as you would expect they would. The only surprises are the platforms on which the contestants stand: They can be raised independently up to about a foot above the level of the stage. This is so that the heads of all three contestants can be made to be roughly level, which makes the job of the person who pans the camera to the responding contestant much easier. As I am 6'4" tall, and thus taller than every single other contestant in both groups, my box stayed permanently at floor level.

The contestants who are not playing sit in the second and third rows of the (stage) leftmost bank of seats; the first row is reserved for those with disabilities. This is where the non-playing folk sat during practice as well.

The practice rounds were basically a reenactment of a full Jeopardy! episode, but with much easier questions and Alex replaced by Glenn (who would do just fine as a game show host himself). Three players, including the returning champion, are invited up to the stage at the beginning of the practice game, and Maggie rotates in new players every few questions. Each player got two chances at the podium. I did very poorly in these practices, but my discussion of buzzer performance will wait for my next post.

The lineup for the first game was announced following the practice, and I was pleased to discover that I wasn't in it. I was still nervous and unsettled, and my weak performance in practice was on my mind. I could tell that I would not have played well had I been put out there.

So, it was off to the green room to cool our heels until the audience could be brought in. We were reminded not to attempt any communication with any audience member, then we were brought in to sit down in our rows. I had to spend the day not making eye contact with my parents, who were only a few seats away.

Watching the game from the seats was much like watching it from home. Instead of commercials, we listened to Alex take questions from the audience, and occasionally, redub a line that he had missed.

The first game was a well-fought battle that saw two-time champ Jericho fall to Sam. After it ended, the two players chosen for the next game (not me) were revealed and we all went back to the green room for refreshments.

As the day passed, the pattern repeated, with the contestant rows slowly becoming depopulated. Lindsey lost a heartbreaker to Sam, then Sam was dethroned by Tim.

After the third game, the audience (excepting the families of potential players) was dismissed. We went off to the commissary for a late lunch, and upon returning, we were given another round of practice. It was here that I figured out my buzzer technique, but more on that in my next post.

I was ready to play, now. I felt good about my technique, my nerves had mostly burned off, and I was ready to go. Unfortunately, I did not get chosen.

The pattern of games repeated twice more with me on the sidelines and a (mostly) new audience. The fourth game of the day saw Mary Anna put on a Jeopardy! clinic and win in an impressive lock, then she had a tough loss to Amy to end the day.

Two of us were left in the contestant pool at the end. We were told we'd play the next day for sure, and probably early in the day. So, it was back to the hotel for dinner with my parents and another trip into the contestant pool the next day.

08 June 2011

My Jeopardy! Story: Prologue

I have thought about going on Jeopardy! since I was in high school. Out of some mixture of laziness and fear, though, I didn't try to audition for the show for many years. I always felt like I might do pretty well, though.

That changed in January of 2009, when Jon Hsu, an old friend and Knowledge Bowl teammate from high school, posted a link to the online test to be a Jeopardy! contestant. I was scheduled to be at a professional conference on the day the test was being offered, so I had the honor of taking it from a room at the San Jose Airport La Quinta Inn. (My brother says that "La Quinta" is Spanish for, "Next to the Denny's;" in this particular case, it meant, "Next to the Burger King.") I believe I got something like 40-42 out of 50 questions right, but I was not chosen for an in-person audition.

I took the test again in 2010, answering roughly the same number of questions correctly, and was invited to audition in Washington, DC, on 27 May. On that day, my wife, saint among mortals that she is, loaded both myself and our then-newborn son into the car and drove into the teeth of downtown DC traffic to take me to my audition.

I was a member of an audition group of 40, one of several such groups convened in Washington, DC that week. After waiting in the hall for a bit, we had our photos taken by Robert, one of the contestant coordinators, and we were shuffled the room for our audition.

The audition itself was presided over by another coordinator, Maggie Speak, who is worth the price of admission all by herself. Much has been written about her hyperkinetic and slightly unhinged presentation style elsewhere, but she must truly be witnessed in person to be fully appreciated.

We took another 50-question test, and once it was collected, we were allowed to compare notes on it. Based on these discussions, I figured that I scored about a 45.

It was then time to demonstrate that we aren't completely hopeless on camera. Maggie explained to us that they were looking for people who could follow instructions, handle the buzzer timing, and answer clearly and boldly. She said they also wanted people who looked like they were having fun, not hyper-intense types who looked like they were coming ashore under fire at Iwo Jima. (Based on my later experiences, all I can say is that I must have been having a particularly good day in that regard.) So, we played some practice Jeopardy! in groups of three, then we each chatted with Maggie as if she were Alex doing the "awkward chat with contestants" portion of the show.

I was in the first group of three to step up to the plate, and was assigned the middle buzzer. The gentleman to my left was a tall, bearded extrovert from, I believe, Texas; a series of transportation mishaps resulted in him making it to the audition by the skin of his teeth. He was so personable, outgoing, and unrattled that I figured he was a shoo-in to make the show if he didn't suck on the written test. (I don't remember a great deal about the woman to my right except that she was fairly quiet.) The practice session went very well for me. I buzzed quickly and often, answering several clues correctly and earning praise from Maggie for being loud and clear with my responses. The chat portion was a lot harder, and I felt like I put on no more than a passable performance.

I sat in the "audience" for the rest of the audition as more folks were put through their paces. There were a couple of people who were confident and calm, many more that were clearly overcoming some nerves, a few people that were far too quiet and meek, and a couple of people that should probably seek some sort of treatment for their Asperger's syndrome. (No offense intended to those with Asperger's and the like; I'm not being flip when I say that these people should probably be seeing a professional.) Comparing myself to this competition, I figured myself to rank about fifth, plus or minus three, out of 40. I thought I had an outside shot at best of making the show.

By January of this year, I had given up hope.

I was wrong. While in Denver International Airport on my way to a conference (the same one where I had taken my first online test two years earlier), I got a call from Glenn, the contestant coordinator. He told me that I was going to be on Jeopardy! It was all I could do not dance in the terminal.

12 May 2011

Statistics and Intuition

Statistical analysis sometimes produces results that many people find hard to accept. I have seen that with my Jeopardy! Tournament of Champions odds calculations, and it seems to pop up in many other places as well.

Recently, there has been a flare-up of discussion about the famous "hot hand" study done 26 years ago. I don't give a damn about spectator sports anymore (and the NBA was very far from my favorite league even when I did), but I find the study fascinating. Briefly, there is an abiding perception among both players and spectators that a player who has hit several consecutive shots is temporarily "hot" or "dialed in," and should thus be given the opportunity to make more shots while still in this state. In statistical terms, people believe that successful shots are correlated, so the probability of a player sinking their nth shot after hitting (n - 1) consecutive shots immediately prior is higher for n > 1 than it is for n = 1 or n = 0. The results of the very careful and thorough study to which I linked above is that this is not correct. To a high degree of certainty, the probability of sinking several shots in a row is pretty much what you'd predict from the player's shooting percentage and the assumption that the shots are independent. Basically, a player with a shooting percentage of 50% will hit five shots in a row just as often as someone flipping a fair coin will have it come up heads five times in a row (on average, once for every 32 such five-event sequences).

So, question settled, right? Not quite; some people simply refuse to accept the results of this study. They look for objections on methodological grounds, but for the most part, they simply refuse to place the results of careful analysis above their own intuitions. It's hard to deny your intuition, even when you know you that you should. The latter article links to a highly speculative "hypothesis" that confirmation bias (and other serious defects of reasoning) are the result of social evolution. I am highly skeptical of this idea, since, like most of the output of evolutionary psychology, it's not a properly falsifiable and testable scientific hypothesis. Whatever the actual causes, though, it seems pretty clear that the reproductive fitness our naked prairie ape ancestors was not impeded by confirmation bias, so we're stuck with it. As was illustrated in the article above, confirmation bias can be so strong that no amount of evidence will be sufficient to override it on some level.

Knowing all of this, I have worked very hard to attack my own confirmation bias. If a new idea seems immediately right to me, I try to adopt the stance that it is invalid, and that I will need to be presented with significant evidence to convince me otherwise. This has led to a certain contrarian confirmation bias in my personality, where the innate appeal of an idea to me varies inversely as its popularity: Every time a pop intellectual proclaims some supposedly-revolutionary idea, the knob on my bullshit filter goes right on past "healthy skepticism" and doesn't stop until "scornful dismissal;" I'm overcorrecting for the large number of otherwise-intelligent people whose knobs never seem to leave "awed, uncritical acceptance" when it comes to Michael Pollan or Malcom Gladwell or the Freakonomics guys. So, even my efforts to defeat my confirmation bias are marred by it!

Whatever internal battles I may wage, though, I am always willing to supplant my flawed intuitions with real knowledge good data and good analysis. It's not always easy to use the real knowledge when my intuition is screaming at me to ignore it, but, as both a scientist and a person, I feel it's vitally important to try.

27 April 2011

Yet Another ToC Odds Update

Wow. It's time for yet another ToC update, since Brian Meacham just ended his very excellent run of four wins.

So, without further ado, on to the table:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,41199.9%100.0%-99.6%
Kara Spak5 Wins$83,40199.4%100.0%-96.5%
John Krizel4 Wins$105,20496.3%99.6%-86.8%
Brian Meacham4 Wins$90,50081.6%95.6%-60.5%
Buddy Wright4 Wins$88,80462.7%86.2%-38.8%
Sara Heard4 Wins$85,60141.2%69.3%-21.0%
Paul Wampler4 Wins$72,00130.5%58.7%-13.5%
Anthony Fox4 Wins$51,9983.2%10.9%-0.9%
Megan Barnes3 Wins$103,2030.3%*1.6%-0.1%
Alison Stone Roberg3 Wins$85,1020.3%1.4%-0.1%

We have had an improbable pile of ToC-grade contestants over the past few weeks. We've had five ToC-grade players in the last 31 games, for a generation rate of about one per 6 episodes. The normal generation rate for players of Megan Barnes' accomplishment level or better is about one per 17 games. So, we're somewhere between two and three standard deviations from the mean here. In other words: This has been a pretty darn weird late winter and early spring.

25 April 2011

Color

"Diagonal and Door," Taken in Greenbelt, MD, 2010 (Sculpture by Lenore Thomas)

I finish far fewer of the photos I take in color than I do in monochrome, and those that I do finish in color, I prefer them to have a limited palette. I find that I understand photos a great deal more in terms of shapes and tones than I do colors, so this is a more natural way of working for me.

The photo above (the first print of which now hangs in my father-in-law's office) is a good example of how I generally like to use color: The red door and the green bush are the only colors with any real saturation, and those colors only serve to highlight the strong diagonal line and the little details on the pure black and white of the sculpture. The photo below provides an even more extreme example, with blue and purple its only colors and the itself completely in the shadow.

"Tree and Skylight," Taken in Greenbelt, MD, 2010


When I do a full color photo, I sometimes feel like I'm fighting against the color rather than working with it. I'll think that the color looks bland and washed out, so I'll jack up the contrast only to make something that loooks garish and overdone. Hues that looked good in the real world appear to clash on the screen. And just when I think I've got everything balanced just so, the print will arrive looking just a bit off from where I want it. (Note to self: Get a better monitor.)

Sometimes, though, I get a full-color photo I like. I highlighed "Ferris Wheel in Morning Light" in an earlier post; below is another of my favorites. I took it last year while at Shenandoah National Park with the intent of converting it to B&W, but I decided I liked the color better. Is it too saturated and over-the top? Maybe, maybe not, but this is how the scene felt. I'm happy with the result.

"A Tree by a Brook," Taken in Shenandoah Park, VA, 2010

20 April 2011

Jeopardy! ToC Odds: New Update

The table below reflects four changes since the last update: First, Sara Heard finished an excellent run of four wins last night; congratulations, Sara! Second, Megan Barnes, in the thread for the previous rankings, announced that she was pregnant and will be unavailable for the ToC if she is selected. So, I have taken her out of the data for this year, but since I am convinced that the producers will invite her to the next ToC if she would qualify for this one but can't be there for it, I'll continue to calculate her odds. Thirdly, I have enlarged the error range to two standard deviations, meaning that there is a 95% chance that the "real" odds of making the ToC for any given player lie within the reported error range.

So, on with the table:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,41199.9%100.0%-99.4%
Kara Spak5 Wins$83,40199.2%100.0%-95.2%
John Krizel4 Wins$105,20495.1%99.5%-83.4%
Buddy Wright4 Wins$88,80475.8%93.5%-52.2%
Sara Heard4 Wins$85,60157.4%83.3%-33.2%
Paul Wampler4 Wins$72,00130.5%58.7%-13.5%
Anthony Fox4 Wins$51,9988.7%24.7%-2.8%
Megan Barnes3 Wins$103,2031.8%*7.0%-0.4%
Alison Stone Roberg3 Wins$85,1021.6%6.5%-0.4%
Sarah Wilkinson3 Wins$72,7010.1%0.6%-0.1%

So, the cutline runs right through our new contender, Sara Heard, with the most likely field featuring her and everyone above her, as well as four people who have yet to play, with Paul Wampler as the alternate. It's interesting to note that even with 95% error bars, the cutline only moves one player in either direction at the extremes. It's also interesting to think that there is a roughly 1 in 20 chance that the Tournament field will be composed entirely of 5+-time winners, and that there is roughly a 1 in 50 chance that we will see only one new ToC contender in the next 100 games or so.

12 April 2011

Jeopardy! ToC Odds: Now With Error Bars!

Christopher Short (cshort on the J! board) ended his run on Jeopardy! yesterday, leaving as a 6-time champion; congratulations, Chris!. This means that it's time to update the Tournament of Champions standings. One improvement that I'll be putting in the table today, at the request of RCraig (a.k.a Roger Craig, also a 6-time champion), is an error range for each estimate.

The data set I use to make my estimates is a compilation of all 3+-time winners since the 2006 ToC, covering the 2007, 2009, 2010, and (upcoming) 2011 Tournaments. As I explained in my previous post, for each champion in this year's field, I use this data set to estimate the number of players who are likely to appear before the next ToC that would be ranked more highly than the player in question.

If the data set I used to make my estimates was infinite (and if the distribution didn't change with time), I would know the probability of generating a player above a given level of accomplishment exactly, and no error bars would be needed. Right now, my data set has 99 listings, which is a long way from infinite, I'm afraid; this means that I know the relevant probabilities imperfectly. To estimate the uncertainty in a given probability, I will once again turn to the Poisson distribution. As you may remember, a convenient feature of the Poisson distribution is that its standard deviation is the square root of its mean. So, if my dataset contains n entries above the accomplishment level in question, the uncertainty in that value is the square root of n. This isn't to say that I'm unsure of the actual data, since even someone like me with a math degree can count; the uncertainty here is in how well the actual data reflects the underlying distribution. (That was a terribly frequentist way of putting it; I'm sure a hardcore Bayesian would disagree with my wording.) I can then feed the values at one standard deviation's remove from the actual count into my estimation method and thus generate error bars.

This might be clearer with an example: Anthony Fox (ARF!) won four times and earned $51,998. Of the 99 champions in my database, 48 are ranked more highly, and I estimate his chances at the ToC at 17.4% based on this. The square root of 48 is ~6.93, so I ran my estimation algorithm using values of 42.07 and 54.93 to give probability estimates of 27.5% and 10.4%, respectively. So, I'd report that my estimate of Anthony making the ToC is 17.4%, with an error range of 27.5% to 10.4%.

It is worth noting that these error estimates reflect one standard deviation of uncertainty; that is to say, there is a ~68% chance that the real chance of a given player lies within my reported error range. One could just as well chose to report an error range of two standard deviations, (41.5% to 6.2% for the above example), which would be ~95% likely to encompass the real probability.

So, with those formalities out of the way, on to the table:

Competitor   # of Wins   Money Won   ToC Chance   Error Range   
Tom Nissley8 Wins$235,405100.0%100.0%-100.0%
Roger Craig6 Wins$230,200100.0%100.0%-100.0%
Christopher Short6 Wins$94,752100.0%100.0%-100.0%
Tom Kunzen5 Wins$133,402100.0%100.0%-100.0%
Paul Kursky5 Wins$109,41199.9%100.0%-99.1%
Kara Spak5 Wins$83,40198.8%99.7%-96.9%
John Krizel4 Wins$105,20493.6%97.4%-87.6%
Buddy Wright4 Wins$88,80471.3%82.9%-58.7%
Paul Wampler4 Wins$72,00145.5%60.0%-32.3%
Anthony Fox4 Wins$51,99817.4%27.5%-10.6%
Megan Barnes3 Wins$103,2035.2%9.7%-2.7%
Alison Stone Roberg3 Wins$85,1021.2%2.5%-0.5%
Sarah Wilkinson3 Wins$72,7010.1%0.2%-0.0%

Based on these, the most likely field for the ToC at the moment includes all of the 5+-time winners, John Krizel, Buddy Wright, and 5 people who are yet to air or yet to play, with Paul Wampler as the alternate.

Note that the estimates for players on the bubble are extremely sensitive to uncertainty in the data used for the prediction, while the sensitivity is much lower when one moves to the extremes.

It's interesting to see that with ~110 games (out of 272) still left to play, we have already converged to having only 3 players with probabilities between 10% and 90%.

A final note about the results: As Robert K S pointed out on my last post, it's entirely possible, based on history, that a 5-time champion who somehow doesn't make this year's ToC will be invited to the next ToC. So, the probabilities for the 5-timers should be viewed as the probability that they will make the 2011 ToC.

08 April 2011

"Dead Tree," Taken in Greenbelt, MD, August 2010
Photography writers often say that a particular picture cannot be fully appreciated on the web, and that you really need to see the print. I was initially a bit skeptical of this statement. The prints under discussion were often smaller than a good-sized, high-resolution computer monitor that could surely display all the detail needed to get a good sense of the picture, right?

My print of the photograph above, "Dead Tree," is what well and truly convinced me otherwise. I was quite happy with the composition and mood of the photograph when I worked it up, with the creepy vibe from the tree stump and its dead branch reaching up like a claw. I always thought, though, that the truly vast number of individual leaves in the shot made the background look busy. When I got my 10.5" x 14" print from my printer, I was shocked at the result: The background was beautiful, adding to the picture rather than detracting from it. I suspect that the impression of busy-ness conveyed by the computer image was, in part, the result of a lot of adjacent pixels with a large difference in tone; the resulting sharp, square edges made the background look displeasing. In the print, though, more detail was resolved and the transitions between tones were less harsh, producing a smoothly detailed sea of leaves.

Although I have not yet had it printed (buying a new house and moving wreaks havoc on time, organization, and budgets), I suspect the photograph below, "Bridge Over Frozen Water," will show a similar improvement upon printing, with the busy clouds becoming smooth and appealing.

So, I'm convinced: If you want to really appreciate a picture, especially those with lots and lots of high-contrast detail, you need to see the print.

"Bridge Over Frozen Water," Taken in Rosslyn, VA, January 2011.
For more of my photographs, please see my photography website.

On the Awfulness of Pandora's Algorithm

I admit that the title of this post is not fair. In many ways, Pandora is a wonderful service, but it can be quite vexing.

For those not in the know, Pandora offers the user customized "radio stations" that play songs that are supposed to be similar to the songs of the artist or artists that the user selects to "seed" the station. So, when I wanted a station that played metal of the thrash, progressive, and death varieties, I seeded it with Slayer, Pantera, Gojira, Mastodon, and Opeth. When I wanted a station that played post-rock, I seeded it with Sigur Ros, Godspeed You Black Emperor!, Explosions in the Sky, and fly pan am. (Observed trend: post-rock bands have much longer and more pretentious names, on average, than metal bands.) You can give each song a "thumbs up" or "thumbs down" rating in an effort to improve the station and play more of the music you like.

Pandora picks songs using something they call the Music Genome Project:
Together we set out to capture the essence of music at the most fundamental level. We ended up assembling literally hundreds of musical attributes or "genes" into a very large Music Genome. Taken together these genes capture the unique and magical musical identity of a song - everything from melody, harmony and rhythm, to instrumentation, orchestration, arrangement, lyrics, and of course the rich world of singing and vocal harmony. It's not about what a band looks like, or what genre they supposedly belong to, or about who buys their records - it's about what each individual song sounds like.
Sounds great, right? They use this comprehensive database of song similarity to pick out your tastes and give you songs that match them, and you happily listen to a radio station that sounds like the Platonic form of your ideal .mp3 player on shuffle.

Their methods works pretty well for the two stations I mentioned above. There is a certain unfortunate tendency of the metal station to play "dinosaur metal," like Iron Maiden, and post-rock station will play electronica from time to time, but they're mostly pretty good.

The problem I've had is with the station that I called "Modern Alternative." My goal was to make a station that ranged across a broad swath of my musical tastes, playing lots of different music from the last two decades that I would enjoy. To get the desired breadth and variety, I seeded it with bands ranging from Silversun Pickups to Smashing Pumpkins to The Dresden Dolls to Ivy. I've listened to the station for many hours, rating hundreds of songs. The result is a complete mess.

Sometimes, it does well: "Portishead and Soundgarden were good calls, little algorithm."

Sometimes it misses, but in completely understandable ways: "Well, I can see where you'd try The Bravery and Bush, but they aren't my favorites."

Sometime, it skews too far towards older music: "I love Led Zeppelin, but that belongs on my classic rock station, not here."

Sometimes it gets fixated on one artist...: "Okay, I just don't like Death Cab For Cutie enough to hear eight songs by them in two hours."

...or on one aesthetic sensibility: "A Talking Heads song, two Modest Mouse songs, and 'I Am the Walrus' is just a little too much po-mo for one morning."

Somtimes it misses horribly, with gruesome results: "Wilco and Johnny freaking bloody buggery Cash? It's like you don't even know me at all!"

At least it hasn't tried to play Snoop Dogg or the Spice Girls yet...

The inability of this station to pick good songs makes me question the merits of their database and algorithm. My metal and post-rock stations, where Pandora does well, are entirely obvious (a brainless genre-tagging method could predict that someone who likes Explosions in the Sky might like Mogwai), and it completely drops the ball when it's asked to do something harder.

Perhaps the take-home message here is that music tastes are too individualistic and dependent on extrinsic factors (like the fact that you heard a given song on the way to your first date with the woman you'd eventually marry, or that another song was played to death by the radio when you were in college) for any computer to predict.

Or maybe Pandora desperately needs to rewrite its algorithm so that my station never again plays Tom Petty (blech).

07 April 2011

More on Probabilities and the ToC

As we have a new member of the 4+ winners' club, Christopher Short, I will be revising my ToC probability estimates... errr... shortly.

It was asked on the Jeopardy! board why I would need to make this sort of revision. After all, didn't my initial estimates take this sort of thing into account. Why would they need to be revised?

The probabilities shift because more information has been added. Every game that passes without adding a 4+-game winner increases the chances of the high-ranked 3-timers making the ToC, while every 4+-game winner added to the pool decreases their chances. While I'm using a strictly Bayesean analysis here (really, the whole "Bayesian vs. Frequentist" debate goes right over my head a lot of the time), the idea is the same: As information is added, the probabilities will change to reflect the new information. If we get a long drought of big winners, then the intermediate 4-timers will have their chances pushed up toward 100% and the highest 3-timers will move up into the middle. If a bunch of 5-timers are suddenly generated, then all the 3-timers will be pushed down below 1% and the intermediate 4-timers will head toward the middle.

Think of it like a game of Jeopardy!: If you know nothing about the 3 contestants at the start of the game, you'd estimate each to have a 33% chance of winning. By the end of SJ!, their scores will have given you information about both the contestants and the game, and you can likely produce a more refined estimate. By the end of DJ!, the estimates will be a lot firmer, with the leader standing the best chance of winning and the two lower-scoring players having concomitantly lower chances. (In the case of lock games, they'll already have converged to 100%/0%/0%.) Finally, as the FJ! answers and wagers are revealed, the estimates of winning can be improved further, gaining accuracy and certainty, until the last wager is displayed and the matter leaves the realm of probability for that of fact.

The closer we get to the actual ToC, and the more "predicted" territory becomes "actual," the more the odds will tend to converge toward 100% or 0%. I'd guess that at the beginning of October, there will be about 11 people who are 95%+ likely to make the tournament and about 3 people with intermediate-to-poor chances, with the rest of the field entirely out of contention.